Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Journal of International Scientific Publication: Educational Alternatives, Volume 9. INTELLIGENT SYSTEM OF KNOWLEDGE CONTROL FOR E-LEARNING Dauren I. Kabenov1, Raihan Muratkhan1, Dina Zh. Satybaldina1, Bibigul Sh. Razahova1, and Altynbek A. Sharipbayev1 1 L. Gumilyov Eurasian National University, 5 Munaitpasov str., 010000 Astana, Kazakhstan Abstract This paper proposes a framework for the design and development of an intellectual information system of knowledge control. Traditional tests application is supposed to choose the answers on the basis of two-positional logic. It is supposed to use them adequately only in terms of strictly formally asked questions. It leads to absolutely simple questions. But knowledge appropriation includes not only (and not so much) memorizing a priori veritable facts but the capability to understanding general phenomena, tendencies. To control this knowledge “open” (without answer variants) test tasks are more effective. In this paper we have presented an ontology- based text mining approach for the automatic evaluation of the student’s answers on the Natural Language (Kazakh language). The use of intelligent algorithms can also dynamically change the system of evaluation and the test circuit, which significantly improves the quality and speed of testing. Keywords: intellectual system, ontology, text mining, control of knowledge, testing systems 1 Introduction Successful e-learning takes place in an integrated system that combines a environment tools for creation of electronic educational materials, subsystems for control of knowledge and subsystems for support of the training process. Also practicing credit-module system of the learning process organization makes working out effective means of students' knowledge control important. The research analysis of this problem shows the tendency of enlarging tests usage as an instrument of the studied material quality evaluation [1]. The main advantage of computer tests is the opportunity to ask all the students within the assignment in equal conditions and according to the equal grades scale. It increases the objectivity of knowledge control in comparison with the traditional methods. Currently, there are many testing systems in various fields of knowledge, for example OLAT [2], Moodle [3], Sakai [4] and AuthorWare [5]. Most of these tools provide the ability to create multimedia tests, testing for traditional learning and e-learning, saving and transfer of results to the teacher for administration of users and educational groups. On the other hand, using traditional tests suggests choosing answers (and their evaluation correspondingly) on the basis of two-positional logic [1]. It leads to absolutely simple question formulations which “lie on the surface”. To control knowledge “open” questions (without suggested variants of answers) are more effective. However, the adequate automatic verification of answers to the questions of this type is a difficult task. Patterns of the answers in the form of regular expressions are not able to take into account the diversity inherent in the native language. Also the automatic detection of random errors (misprints, typographical errors), and spelling is required. In order to alleviate this disadvantage of test control of knowledge in this paper we suggest an application of artificial intelligence methods and tools, in particular, ontological engineering. 1 Journal of International Scientific Publication: Educational Alternatives, Volume 9. Description of domain knowledge Planimetry ontology is presented in the paper. Classes of concepts, their structure and properties are defined. List of used relations and characteristics of output procedures are considered. For texts of the geometric tasks solution ontological descriptions of the situations presented as a result of the transformation and evaluation of the concept's structures. It is shown that the use of ontologybased text mining can open the "anatomy" of the correct answer preparation that can be used in the analysis of a student's answer and the search for precisely that moment, which caused difficulties in their reasoning. Results of research can be used at creation of intellectual testing systems on the base processing of the Kazakh language. The proposed concept testing system enables the use of intellectual evaluation results of the user level and provides a set of tests, tailored to the level of preparation of the test. The system distributes the issues in terms of complexity, based on data obtained during testing. This enables the construction of adaptive tests, which are self-correcting to the level of users. The paper is organized as follows. Section 2 outlines related work. In section 3 we carry out Planimetry ontology. Section 4 defines the proposed concept testing system based on ontology-based text mining. We finish with some conclusions and future work in section 5. 2 Related Work 2.1 The computer-aided testing of knowledge For many countries, e-learning is valued and utilized as a driving force to speed up the technical, industrial and economic development of the society. As a research subject, e-learning is both multidisciplinary and interdisciplinary and covers a wide range of research topics, with scholars from different disciplines conducting e-learning related research ranging from content design to associated policy. Longitudinal trends of e-learning research using text mining techniques are described in [6]. The authors analyze a significant number of research works and provide useful insights into that elearning research is at the early majority stage and focus has shifted from issues of the effectiveness of e-learning to teaching and learning practices. Current e-learning theories stress the importance of situated cognition and personalized learning [6]. Control of knowledge is an intellectual problem, demanding a high-quality solution that will help to reach a new stage in the methodology of teaching, since it could give the opportunity to realize the idea of individual approach to training on a massive scale. The computer-aided testing of knowledge becomes very popular nowadays, firstly, because it saves the working time of a teacher, relieves him from routine work and allows to provide impartial evaluation of knowledge, the results of which do not depend on the subjective opinion of different teachers. In [7] the newly developed computerized constructive multiple-choice testing system is introduced. The system combines short answer (SA) and multiple-choice (MC) formats by asking examinees to respond to the same question twice, first in the SA format, and then in the MC format. The authors of [8] have developed the software tool that allows to prepare test questions and conduct testing using any of the suggested types of questions below. Description of this software tool and the intellectual algorithms for evaluation of knowledge is presented in the previous paper of the authors [9]. The papers [10, 11] describe methods of implementation of a control mechanism of student knowledge with the help of fuzzy set theory combined with neural network technology. The papers apply some serious improvements in the logic of evaluation of knowledge, and methodologies of data interpretation of student responses. The presented architecture is typical of the configuration of 2 Journal of International Scientific Publication: Educational Alternatives, Volume 9. hardware and software in an intranet environment of educational institutions. Analysis of the aforementioned work shows that ordinary linear tests with simple forms of the answer do not quite meet the requirements of comprehensive control of students' knowledge. Most of all it concerns natural and mathematical sciences, a feature which is the close relationship of concepts, themes and sections of the course, as the main criterion for learning - the ability to solve tasks of different nature and level of complexity. Therefore development of the adaptive, nonlinear, and intellectual testing methods with more different types of tasks and answers' forms are needed. At the same time, new testing systems should incorporate all the achievements of previous generations of the knowledge control tools. 2.2 Ontologies In order to build ontology of Planimetry, it is beneficial to understand the need of ontology and some works concerned of the ontology-based text mining. An ontology is an explicit formal specification of the terms in explicit specification the domain and relations among them [12]. Ontologies are useful as means to support sharing and reutilization of knowledge [13]. This reusability approach is based on the assumption that if a modeling scheme, i.e., ontology, is explicitly specified and mutually agreed upon by the parties involved, and then it is possible to share, reutilize and extend knowledge. Many disciplines now develop standardized ontologies that domain experts can use to share and annotate information in their fields. Problemsolving methods, domain-independent applications, and software agents use ontologies and knowledge bases built on ontologies as data [14]. Reusing existing ontologies may be a requirement if our system needs to interact with other applications that have already committed to particular ontologies or controlled vocabularies [14]. There are libraries of reusable ontologies on the Web and in the literature, for example, the Ontolingua ontology library [15], or the DAML ontology library [16]. The need of ontologies is connected with the inability of the existing methods to adequately automatically process native-language texts. For high-quality word processing, you must have a detailed description of the problem area with a lot of logical links that show the relationships between the terms field. The use of ontologies can provide a native language text in such a way that when it becomes available-for-automatic processing [17]. In this paper we develop the Planimetry ontology and this ontology is used as a basis for the automatic verification of geometric task solution. The authors of [18] have developed the similar ontology for automatic synthesis of structural images of the planimetric figures. We use some concepts of the ontology, obtained the authors consent. 3 Planimetry ontology An ontology is a formal explicit description of concepts in a domain of discourse (classes (sometimes called concepts)), properties of each concept describing various features and attributes of the concept (slots (sometimes called roles or properties)), and restrictions on slots (facets (sometimes called role restrictions)) [14]. Ontology together with a set of individual instances of classes constitutes a knowledge base. Development of an ontology includes [14]: • defining classes in the ontology, 3 Journal of International Scientific Publication: Educational Alternatives, Volume 9. • arranging the classes in a taxonomic (subclass– superclass) hierarchy, • defining slots and describing allowed values for these slots, • filling in the values for slots for instances. Our ontology is structured in three levels. The first level contains classes whose instances cannot be derived from other classes. Class Plane Shape represents all objects of planimetric shapes. Specific Plane Shapes - instances of this class. The next levels are generated through a reasoning process, that is, using the ontology reasoner or through the different modules called by the Reasoning Manager. The higher the level the more detailed the information is, e.g., in the second level polygons are classified in pentagons or triangles or quadrangles and in the third level the triangle are further classified in rectangular, equilateral and isosceles triangle (if the classification by sides will be considered, see Figure 1). Fig. 1. Types of triangles. Classification by sides and classification by angles. We organize the classes into a hierarchical taxonomy by asking if by being an instance of one class, the object will necessarily (i.e., by definition) be an instance of some other class. If a class A is a superclass of class B, then every instance of B is also an instance of A In other words, the class B represents a concept that is a “kind of” A [14]. For example, every equilateral triangle is necessarily a isosceles triangle. Every isosceles triangle is necessarily a triangle. Therefore the equilateral triangle class is a subclass of the equilateral triangle class. Slots describe properties of classes and instances. Each property can be set to a specific value or a formula to calculate this value of the property: for example, a triangle has sides, has angles, sides have lengths, angles have a degree measure. All subclasses of a class inherit the slot of that class. On the other hand, subclasses can have their own properties. For example, an isosceles triangle has sides that have a length, with those two sides (legs) are congruent, and the third party has a special name - base (see Fig.2 and Fig 3). 4 Journal of International Scientific Publication: Educational Alternatives, Volume 9. Fig. 2. Isosceles triangles. The angles opposite the two congruent sides are called base angles and the base angles. The angles opposite the base is called vertex angle. Fig. 3. Structure of the isosceles triangle concept to parse of the solution of the geometric tasks. Fragments of the ontology, including the structure and properties of concepts are the basis for description of the situation, which is determined by the input data to solve a geometrical task. The concepts and relationships defined by the input conditions (the geometrical task's text) are introduced in addition to these ontology fragments. 5 Journal of International Scientific Publication: Educational Alternatives, Volume 9. 4 Proposed concept testing system based on ontology-based text mining Compared to traditional forms of learning, e-learning has several advantages: adapting to the individual characteristics of students, the freedom to choose the time, place and level of education, use of new teaching methods, modern means of communication and information transfer between a student and a teacher. However, control of knowledge is particularly important because of the lack of direct contact between student and teacher. Organization of control of knowledge is closely connected with the problem of selecting the type of questions, mode of the testing trajectory formation and methods of the answers verification. To solve these problems we propose the concept of intellectual testing system on the base of the domain knowledge ontology. Planimetry ontology is used as the domain knowledge ontology. The following types of questions are offered for the control of knowledge quality: - test questions of closed form, i.e., when several variants of the answer are suggested, one of which is correct and should be selected; - test questions of open form, i.e., questions without suggested variants of answers (such questions are useful for evaluation of knowledge of terms, definitions, notions, etc); - situation tests, i.e., set of test assignments designed for solution of problematic situations (a geometrical task). A special method of selection of a testing trajectory mode is proposed. Test set is not formed by random sample from tests database. Questions selection is based on the analysis of answers to previous questions. This algorithm is based on the original method of choosing questions according to the system, corresponding to the current level of student knowledge [11]. For analysis and verification of student's answers to test questions of open form ontology-based text mining is proposed. Description of domain knowledge ontology of Planimetry was presented in Section 3. Semantic analysis of the native language texts is the following stage. 1. Pre-linguistic processing of the source text (morphological and syntactic analysis of sentences) is needed to separate terms (classes, subclasses, properties and relations). 2. A formal understanding of the text as result of constructing an ontological graph. In paper [19] the formalized syntactic rules, analysis and synthesis algorithms of word-combinations and sentences of the Kazakh language were constructed. The results of research can be used at creation of intellectual human-machine systems with interaction possibility in the Kazakh language. Let us describe in brief the proposed method of verification of the geometric task solution on the basis of ontology. Texts of geometrical task are a set of connected sentences [18]. These include simple and complex sentences, incomplete sentences (with an anaphora and an ellipsis). A formal understanding of the text of geometrical problems is their representation in the language of domain knowledge ontology of Planimetry. This representation must be connected and extended with filling in the values for slots for instances from the description of the situation presented by the text. Consider the whole process of analysis with geometric problems in this paper is not possible. Therefore, let us consider the situation's structure which should be obtained as a result of ontologybased linguistic analysis for one geometric task. Text of task: determine the type of triangle that has sides of length which is equal to 5, 6, 6. Pre- linguistic processing of the source text will give the following concepts and combinations of concepts that are comparable to the ontology presented in Figure 3, for example: type (of triangle); 6 Journal of International Scientific Publication: Educational Alternatives, Volume 9. has sides (from triangle class); has length, is equal. For the formal description of Planimetry ontology Prolog (a logic programming language) is used [20]. Prolog's inference engine used to build an ontological graph. A formal description of the ontology (see Fig. 3) uses the following data types with using alternative functors. domains poligon = triangle; pentagon; quadrangle triangle=triangle1(classification_by_sides); triangle2(classification_by_angles); classification_by_sides=scalene; isosceles_triangle; equilateral_triangle scalene=sc(sides) isosceles_triangle=is_tr(sides) equilateral_triangle=eq_tr(sides) sides=sides(side_a, side_b, side_c) side_a, side_b, side_c =real classification_by_angles=acute_triangle; right_triangle; obtuse_triangle acute_triangle=ac_tr(angles) right_triangle=r_tr(angles) obtuse_triangle=o_tr(angles) angles=angles(angleA, angleB, angleC) angleA, angleB, angleC =real predicates nondeterm treangle_tip(treangle,symbol) Hence, an inference rule describing that the triangle is isosceles can be given by treangle_tip(treangle(sides(Side_a,Side_b,Side_c), angles(Angle_A,Angle_B,Angle_C)),Y):-Side_a=Side_b, Angle_A=Angle_B,Y=isosceles; Side_b=Side_c,Angle_B=Angle_C,Y=isosceles; 7 Journal of International Scientific Publication: Educational Alternatives, Volume 9. Side_a=Side_c,Angle_A=Angle_C,Y=isosceles,!. treangle_tip(treangle(sides(Side_a,Side_b,Side_c), angles(Angle_A,Angle_B,Angle_C)),Y):Side_a<>Side_b,Side_b<>Side_c, Angle_A+Angle_B+Angle_C=3.14,Y=scalene,!. treangle_tip(treangle(sides(Side_a,Side_b,Side_c), angles(Angle_A,Angle_B,Angle_C)),Y):Side_a=Side_b, Side_b=Side_c,Y=equilateral; Angle_A=3.14/3,Angle_B=3.14/3, Angle_C=3.14/3, Y=equilateral. treangle_tip(treangle(sides(Side_a,Side_b,Side_c), angles(Angle_A,Angle_B,Angle_C)),Y):Angle_A<1.57,Angle_B<1.57, Angle_C<1.57, Angle_A+Angle_B+Angle_C=3.14,Y=acute. treangle_tip(treangle(sides(Side_a,Side_b,Side_c), angles(Angle_A,Angle_B,Angle_C)),Y):Angle_A=1.57, Angle_A+Angle_B+Angle_C=3.14, Y=right; Angle_B=1.57, Angle_A+Angle_B+Angle_C=3.14, Y=right;Angle_C=1.57, Angle_A+Angle_B+Angle_C=3.14,Y=right. treangle_tip(treangle(sides(Side_a,Side_b,Side_c), angles(Angle_A,Angle_B,Angle_C)),Y):Angle_A>1.57,Angle_A+Angle_B+Angle_C=3.14, Y=obtuse; Angle_B>1.57, Angle_A+Angle_B+Angle_C=3.14,Y=obtuse; Angle_C>1.57,Angle_A+Angle_B+Angle_C=3.14, Y=obtuse. 8 Journal of International Scientific Publication: Educational Alternatives, Volume 9. 5 Conclusions and future work This paper presents an ontology-based approach that addresses the problem of ontology-based text mining. We have developed Planimetry ontology for automatic verification of answers to the questions of “open” questions (without options). Results of the research can be used in creation of intellectual testing systems on the base of the Kazakh language processing. Our conception of the testing system uses intellectual evaluation results of level of the user and provides a set of tests, tailored to the level of preparation of the student. The control knowledge system distributes the issues in terms of complexity, based on the data obtained during testing. This enables the construction of adaptive tests, which are self-correcting to the level of users. Future research in the frame of methodological aspects of computer -aided control knowledge will be concern to development of the test questions databases of different types and different levels of complexity. Also in the technical aspects frame researches will continue development software that implements of all the stages of semantic analysis of texts on the basis of the ontological engineering and native languages processing. ACKNOWLEDGEMENTS: The present work is supported by the Fund of Ministry of Education and Science of the Kazakhstan Republic (grants № 120, "Automation of recognition and generation of the Kazakh written and spoken languages"). References 1. Prisyazhnyuk, E.: Fuzzy model of the Automated System of Module Knowledge Control. Int.l J. Information Technologies and Knowledge, Vol.2 (2008) 465-468. 2. OLAT – The Open Source LMS. [Online] Available from: http://www.olat.org/ website/en/ html/index.html. 3. Moodle Trust. Moodle.org: Open-source Community-based Tools for Learning. [Online] Available from: http://moodle.org/. 4. Sakai Foundation. Sakai Project. [Online] Available from: http://sakaiproject.org/. 5. Adobe Systems Incorporated. Adobe Authorware 7. [Online] Available from:http:// www.adobe.com /products /authorware/ 6. Hung Jui-long.: Trends of E-learning Research from 2000 to 2008: Use of text mining and bibliometrics. British J. of Educational Technology,Vol. 43.1, (2012) 5–16 7. Park, Jooyong.: Constructive multiple-choice testing system. British J. of Educational Technology. Special Issue: Learning objects in progress, Vol. 41.6, (2010) 1054–1064. 8. Alexeyev, A.N., Volkov, N.I., Kochevsky, A.N.: Types of Questions for Computer-Aided Testing of Students’ Knowledge Computer Modelling and New Technologies, Vol.11.3, Riga (2007) 35–42. 9. Alexeyev, A.N., Volkov, N.I., Kochevsky, A.N.: Elements of Fuzzy Logic in Computer-Aided Control of Knowledge. Otkrytoye obrazovaniye, Vol. 4, Moscow (2003) 23–25 (In Russian). 10. Shahbazova, Sh., Freisleben, B.A.: Network-Based Intellectual Information System for Learning and Testing. Proceedings of the Fourth International Conference on Application of Fuzzy Systems and Soft Computing, Siegen, Germany (2000) 308-313. 11. Shahbazova, Sh.: Application of Fuzzy Sets for Control of Students Knowledge. Application 9 Journal of International Scientific Publication: Educational Alternatives, Volume 9. Comput. Math., Vol.10.1, Special Issue (2011) 195-208. 12. Gruber, T.R.: Toward principles for the design of ontologies used for knowledge sharing. International. J. of Human-Computer Studies, Vol.43(5-6), (1195) 907–928. 13. Decker, S., Erdmann, M., Fensel,D. and Studer, D. Ontobroker: Ontology based access to distributed and semi-structured information. DS-8: Semantic Issues in Multimedia Systems, 1999. 14. Noy, Natalya F. and McGuinness, Deborah L.: Ontology Development 101: A Guide to Creating Your First Ontology'. Stanford Knowledge Systems Laboratory Technical Report KSL-01-05 and Stanford Medical Informatics Technical Report SMI-2001-0880, March 2001. 15. Maedche, A., Neumann, G. and Staab, S.: Bootstrapping an Ontology-based Information Extraction System. Intelligent exploration of the web, Physica-Verlag GmbH Heidelberg, Germany, (2003) 345 – 359. 16. The Ontolingua ontology library. http://www.ksl.stanford.edu/software/ontolingua/ 17. [Online] Available from: The DAML ontology library. [Online] Available from: http://www.daml.org/ontologies 18. Khakhalin, G., Kurbatov, S., Naidenova K., Lobzin A.: Integration of the Image and NL-text Analysis/Synthesis Systems. In Rafael Magdalena (eds): Intelligent Data Analysis for Real-Life Applications: Theory and Practice. IGI Global (2012). 19. Razakhova, B.: Automation of the analysis and synthesis of word-combinations and sentences of the Kazakh language. PhD thesis (in Kazakh language). L. Gumilyov Eurasian National University. Astana (2009). 20. Visual Prolog 7.3. [Online] Available from: http://www.visual-prolog.com 10